A daily market update from FS Insight — what you need to know ahead of opening bell

“Really, what I’m aiming for is world domination.” — Alexander McQueen 

Overnight

Foreign investment into China falls to 30-year-low

Treasuries gain, outperform Bunds ahead of 20Y sale; Fed minutes

Amazon joins the Dow Industrials, taking the place of booted-out Walgreens

Walmart’s $2.3 billion purchase of TV maker Vizio is meant to compete against Amazon for ad dollars; bad news for Roku

Arkhouse Management nominates 9 directors to Macy’s BoD, attempting to seize control of the company in continuation of months-long saga

HSBC’s 2023 profit misses estimates; buyback planned

Japan downgrades economic view for first time in three months

Japan’s wage talks closely watched as BOJ mulls rate hike timing

South Africa to dip into contingency reserves to cut debt

Congress announced a new 24-member AI task force, a project that had been a top priority for ex-speaker Kevin McCarthy

Court documents say key witness in impeachment case against Biden was in contact with Russian intelligence

AI Beat

Adobe launches AI assistant that can search and summarize PDFs

NYT plans to debut new generative AI ad tool later this year

AI testing startup Antithesis nabs $47M seed funding at $215M valuation; uses AI to catch software bugs

First news

•  OpenAI piggybacks on Microsoft’s Bing to enter yet another area of Internet-based endeavor.

Charts of the Day

MARKET LEVELS

MORNING INSIGHT

Good morning!

For the 1H top, there are 3 things we are generally looking for and none of these 3 conditions is in place yet. 

Also: tune into our Super Granny webinar, happening today.

Click HERE for more.

TECHNICAL

Due to a medical emergency affecting Mark Newton, our Head of Technical Strategy, there is no technical update this morning. Thank you for your understanding.

CRYPTO

• Ethena, a newly launched stablecoin platform, seeks to offer a yield-bearing synthetic USD product by accepting ETH as collateral. It stakes this collateral and shorts ETH to establish a delta-neutral position, aiming to harvest yield from both staking and the basis paid from positive funding rates. Since its inception, Ethena has minted over $287 million in USDe, drawing attention with its innovative approach that currently offers a 27% annual yield on staked USDe tokens—a rate adjustable weekly. Despite its move from beta to public access and the generation of significant inflows, Ethena has encountered skepticism regarding the viability of its model and risk management strategies, with critics pointing to past failures of similar ventures. A primary concern is the potential for a negative funding regime, where ETH enters backwardation, causing the short ETH trade to negatively impact the collateral value as funding moves in the opposite direction. Although the project has indicated the presence of risk management measures, such as an insurance fund, the real test will come with any significant market downturns.
• Starknet is scheduled to launch its native token, STRK, today, with trading set to commence on centralized exchanges such as Binance, Bybit, Bitfinex, and OKX following the initiation of airdrop claims for over 700 million STRK tokens. This airdrop initiative aims to distribute 7% of Starknet’s total 10 billion token supply across nearly 1.3 million wallets, rewarding early users, Ethereum contributors, and open-source developers. Starknet, an Ethereum Layer-2 network that leverages ZK-Rollup technology, intends to use STRK to decentralize governance. The plan to unlock over 1.31 billion STRK for core contributors and investors soon after launch has drawn criticism, mostly due to the pre-launch vesting schedule provided to early contributors. Starknet is expected to achieve a fully diluted valuation of over $20 billion, positioning it above both ARB -6.39% and OP -2.90%, the two Layer-2 networks that currently dominate L2 activity. Given that Starknet’s daily transaction volume was 80% lower than Arbitrum’s and more than 60% below Optimism’s just yesterday, this valuation has indeed raised some eyebrows.

FIRST NEWS

All in the Family. When Russia went ahead with its full-scale invasion of Ukraine and Western Europe all but stopped buying oil and gas from the aggressor, there was much handwringing about the trap Europe had gotten itself into by previously relying on a single provider – Russia – for most of its hydrocarbons. In the end, the will to move fast (the U.S. sending LNG shipments to Europe by sea), think creatively (Germany putting up LNG terminals seemingly overnight), and leverage a bit of luck (a warmer-than-expected winter meant Germany didn’t have to throttle back its natural-gas-powered manufacturing sector by all that much) meant that a near-monopoly was broken with minimal ill effects, while a way forward toward a multi-provider future was delineated.

There are lessons here for the near-total dependence of AI-providers on chips made by Nvidia. Of course, no one is comparing Jensen Huang to Vladimir Putin or saying that Nvidia is attempting to weaponize its dominant position in the market for AI chips, but the very fact that there are only so many of these chips to go around means that customers need to adjust to Nvidia’s manufacturing schedule and prior commitments. The larger the client, the less they are willing to make those adjustments in the medium to long term, and the more will, wherewithal, and creativity they can muster to leap the paradigm. This is what Microsoft is doing with its latest step taken to reduce dependence on Nvidia.

It may take OpenAI a while to raise the trillions Sam Altman says are necessary to set up new manufacturers of AI chips, but its partner Microsoft is already doing what it can to unwind its reliance on Nvidia by developing its own networking card (used to ensure that data moves quickly between the servers responsible for AI) so as to phase out the one supplied by Nvidia. Any advantage over competitor Google, which also relies on Nvidia chips, is perceived as being worth the extra effort.

Tunneling through the mountain toward the same goal of getting the better of Google, Microsoft’s little nephew OpenAI has been developing ‘a web search product’, partly powered by Microsoft’s Bing search engine, that would bring the Microsoft-backed startup into more direct competition with Google. The move to launch a search app comes a year after Microsoft CEO Satya Nadella said his company would “make Google dance” by incorporating OpenAI’s – well – AI into Bing. That partnership has failed to dent Google’s search dominance so far.

It isn’t clear whether the search product would be separate from OpenAI’s ChatGPT chatbot, which also uses Bing’s index of the web to answer some questions. It could be that OpenAI wants to speed up the service, which can be slow because it also performs tasks like proofreading and summarizing documents. The inherent challenge is two-fold: 1.  the disconnect between the manner in which large language models work (they’re trained on huge, static blocks of data, not a steady stream of new information, the way a search engine, which indexes the web constantly, is) and 2. the time-sensitivity of search – relying as it does on incorporating new information that LLMs aren’t built for. Bing first injected ChatGPT into its search engine a year ago, but that attempt failed to move the needle on moving search audiences away from Google (87.5% share of the U.S search market) toward Bing (7.8% share). Marketing campaigns on Bing do offer lower cost-per-click rates, but Google still commands 80% of the money spent on such campaigns.

Still, the sheer size of those monopolistic numbers makes the target too tempting not to attempt a takedown. With Copilot joining Bing, newcomers emerging (think Perplexity and its cousins) and Google’s Gemini Copilot trying to call Microsoft’s AI wildcard, the landscape of search competition is becoming more crowded than it’s been in at least a decade.

The true wildcard may be how the little nephew  – who, for the moment, needs the allowance from uncle (the $10 billion investment*, for operations) and the ice-cream money (the cloud-computing credits; sugar for that LLM brain) but likely has his own designs on the throne – yes, as we were saying: the true wildcard may be how OpenAI and its ‘web search product’ (let’s call it a search engine out of a sense of the old-fashioned) might leverage Microsoft’s Bing for certain types of functionality.

Industry watchers say they would color themselves surprised if a search interface from OpenAI, especially one relying on the much less popular web search tool that is Bing – whose name, when it debuted, was immediately decoded as “But it’s not Google” – actually made a dent in Google’s search dominance. Still, if OpenAI’s history is any indicator, surprises are in its genes. ChatGPT’s rate of adoption was certainly a surprise, and last week’s debut of Sora was a mouth-agape moment. If OpenAI has hoarded enough lessons that it could apply to search in a (productively) disruptive way, Google could be in for a rude awakening.

All of this may or may not be why Nvidia stock is down lately (likely not) but it is clear that before a single brick is laid down on a factory producing Microsoft/OpenAI silicon, the non-nuclear family members** will exploit every extant way of not using Nvidia chips to get done what they need to be done. Slashdot, The Information, Tech Startups, AdWeek

^{* Although, as we wrote yesterday, OpenAI has just completed a tender offer that gives it more cash at a valuation of $80 billion.}

^{** Although, as we wrote back in December (Grave New World), Microsoft and OpenAI are very much hoping to go nuclear, with Sam Altman’s portfolio company Helion working on a fusion process that has been hitting some early milestones and may start generating energy in a few years.}